IMDHMIControls LibraryShop main store

Explainer

What an EtherNet/IP IO-Link master exposes to your PLC

Industrial networks Published 7 min read

Short answer

Full access to every port is decided by three documented items: the mode each port is configured in, the master's EtherNet/IP data map, and your PLC's client functions. Process data travels on the cyclic I/O connection; parameters need explicit CIP requests, and at 100 devices polled every 10 ms you would need 10,000 transactions per second, so bulk data belongs on the assembly connection.

On this page
  1. What "full port access" means
  2. What each port mode exposes
  3. The two data paths over EtherNet/IP
  4. Reading the master's data map
  5. Parameters are a separate service
  6. How far explicit messaging scales
  7. The decision for each case
  8. Questions asked

Connecting an IO-Link master to your PLC over EtherNet/IP gives you exactly what three documented items allow: the operating mode of each port, the master's EtherNet/IP data map, and the client functions in your controller. The cable guarantees reachability, not exposure. Cyclic process data from every IO-Link port arrives through the master's input and output assemblies on an I/O connection; parameters arrive one transaction at a time through explicit CIP requests. So the question of whether you get full access to the input and output functionality of each programmable port resolves to yes when all three items line up, and to a partial map when any one of them does not.

What "full port access" means #

A programmable port is a port whose behavior you choose in the master's configuration tool, not a port that does everything simultaneously. IO-Link is specified in IEC 61131-9, which defines four operating modes for a master port: IO-Link communication, digital input, digital output, and deactivated. Full access therefore means every port that needs smart data runs in IO-Link mode, its process data appears in the cyclic assemblies, its status bits travel beside it, and its parameters are reachable through the master's documented parameter service.

What the port can never do is add capability the connected device lacks. Process-data direction belongs to the device: a distance sensor supplies input data only, and connecting it to a programmable port creates no output words for it. An actuator consumes output data and usually returns input data such as position or fault flags.

None of these operating modes is a safety function. Light curtains, E-stops and guard door interlocks stay on the safety-rated wiring and controllers your risk assessment calls for, not on IO-Link process data.

What each port mode exposes #

Port mode Cyclic data at the PLC Parameter access Use it for
IO-Link Process data in and out as the device defines, up to 32 bytes each direction under IO-Link v1.1, plus per-port status bits Yes, through the parameter service Smart sensors, valve terminals, RFID heads
Digital input One input bit per port No Dry contacts and 24 V sensors without IO-Link
Digital output One output bit per port No Simple 24 V actuators
Deactivated None No Unused ports

The mode is exclusive per port. A port set to digital output will not speak IO-Link, and a port set to IO-Link is not simultaneously a plain digital channel even when its process data is one bit wide. Selecting a mode changes what the port does; it does not expose every IO-Link function on the same pins at once.

The two data paths over EtherNet/IP #

Implicit messaging is the I/O connection. Your PLC is the scanner, the master is the adapter, and both sides move a fixed block of assembly bytes at the requested packet interval you configure, with no per-point requests in your program. The input assemblies carry process data and status from every active port; the output assembly carries output process data to every port that has any. One connection carries the whole master.

Explicit messaging is the client/server path. Your program sends one CIP request and waits for one response, over TCP. It is the right tool for parameters, identification data, configuration and diagnostics: data that changes rarely and does not belong on a cyclic connection.

Implicit I/O connection Explicit CIP message
What moves Fixed assembly bytes, all ports One service request and response
Pace Every RPI, set at the connection Whenever your code fires it
Carries Process data, status bits Parameters, identification, configuration, diagnostics
Roles PLC is scanner and owner, master is adapter PLC is client, master is server
Load One connection, no per-point code One round trip per request

Support for one path does not grant the other. A controller that can send explicit requests is not automatically able to originate the I/O connection, and a successful explicit identity read proves nothing about cyclic data. Writing outputs by aiming explicit messages at assembly bytes instead of through the connection that owns them is a documented dead end on other adapters as well, the failure mode walked through in why explicit writes to Omron assemblies fail over EtherNet/IP.

Reading the master's data map #

The data map is a vendor document, published as a manual section and as the EDS file your engineering tool reads. It states which assembly instance carries which bytes, the offset of each port's data inside that instance, the layout of the status bits, and the CIP objects behind parameter access. Program against those offsets, not against guesses: check byte order, signed versus unsigned fields and bit-packing before you trust a value in control logic, and prove the interpretation against a known device state.

One documented mapping to anchor this: ifm's AL1920 IO-Link master, configured for ControlLogix, lands port data in the controller tags _Master:I1.Data for inputs and _Master:O1.Data for outputs, and ifm publishes a 66-page EtherNet/IP connection guide that walks the assembly configuration with an Omron NJ/NX controller. Your master's tag and assembly names will differ; its own documentation is what does not.

The status bits are part of the access you are buying. Per-port communication status and data-validity flags let your logic discard data from a port whose device has dropped off, without waiting for a parameter read to time out. ODVA has an IO-Link integration effort that standardizes master behavior on CIP networks, but the vendor's own map remains the authority for the bytes you program against.

Parameters are a separate service #

Parameters live behind the IO-Link ISDU: indexed storage in the device, reached through the master with one explicit transaction per read or write. These transfers slot in between the port's cyclic exchanges, so they run far slower than process data. Treat them as commissioning actions or slow-poll items, not as live control values. Ranging a sensor, reading a device's serial number or changing a valve characteristic all take this path.

On other fieldbuses the same ISDU storage surfaces differently. On PROFINET it is read as acyclic record data, the route described in reading IO-Link ISDU parameters over PROFINET acyclic record data. The parameters do not change; the wrapping does.

How far explicit messaging scales #

Explicit throughput is the limit engineers actually hit. As the client, your PLC paces its own requests, so the ceiling is transactions per second: how fast your controller issues requests, how fast the targets answer, and what your schedule tolerates. Work the requirement backwards.

Requirement: 100 devices, each refreshed every 10 ms by explicit reads.

  • Demand: 100 requests per refresh interval, 100 intervals per second, so 100 × 100 = 10,000 transactions per second.
  • Per-request budget: the refresh interval divided by 100 = 0.1 ms, or 100 µs, including the TCP round trip and the target's response time.
  • Planning figure: 1 ms per explicit transaction on a quiet switched network, a realistic starting number you replace with your own measurement on one target.
  • Sequential round: 100 × 1 ms = 100 ms, so each device is refreshed 1,000 ms / 100 ms = 10 times per second, one tenth of the requirement.

The decision follows from the arithmetic: no sequential explicit client meets a 100 µs per-transaction budget, so process data goes on the I/O connection, one connection per master with no per-point requests. Explicit traffic then carries parameters: polling every device's parameters once per second is 100 transactions per second, one tenth of the 1,000 per second this example network sustains.

Two failure shapes remain. Requests queued one behind another stall on a single dead target, so set each message's timeout below the polling period and sequence the queue so a timeout cannot block the whole round. And diagnostics polled fast crowd out the parameter writes that matter, so slow the diagnostic reads first when margin runs out.

The decision for each case #

  • Machine-speed process data: the I/O connection with the PLC as owner. Every IO-Link port's data and status arrives in the assemblies at the RPI you set.
  • Actuator control: the output assembly on that same connection. Outputs have one owner; do not point a second controller at the same output data.
  • Parameters and device identification: explicit ISDU reads and writes, on demand or on a once-per-second schedule.
  • Bulk counts such as 100 points at that refresh rate: cyclic assemblies, never explicit polling. The arithmetic above settles it.
  • SCADA or a historian needing the same data: read it from the owning PLC or attach as a non-owner client, the pattern shown in how a SCADA system reads EtherNet/IP adapter sensor data.
  • Plain digital devices: run their ports in digital modes, accept one bit each, and reserve the IO-Link ports for the devices whose parameters you actually need.

Configure the port modes first, wire the assemblies second, and spend explicit bandwidth only where a commissioning step or a diagnostic flag asks for it.

Questions asked

Does an IO-Link master expose register numbers like Modbus TCP?

No. On EtherNet/IP the master is a CIP adapter: process data sits in assembly instances at fixed byte offsets, and parameters are object services, not numbered registers. Many master families, ifm's included, can also run Modbus TCP, PROFINET or EtherCAT, but that is a separate fieldbus configuration of the same hardware with its own addressing, not extra registers beside the CIP data.

Can a second controller or a SCADA package read the same master?

Yes, but only one of them owns the I/O connection. Additional clients attach with explicit reads, or read the data out of the owning PLC instead. The master's manual states how many simultaneous connections it accepts; treat that number as the ceiling before you plan redundant clients, and never point two controllers at the output assembly.

What RPI should I configure for the connection to an IO-Link master?

Start at 10 ms. The master refreshes each IO-Link port on its own millisecond-class cycle, so that RPI delivers essentially fresh port data without loading the network. A faster RPI mostly retransmits unchanged bytes; go below that only when the measured port cycle and the machine's timing genuinely need it.

Can some ports run IO-Link while others run digital modes?

Yes. Mode is set per port, so one master can run IO-Link sensors, digital inputs and digital outputs side by side. Only the IO-Link ports contribute parameter traffic; the digital ports appear as single bits in the same assemblies, at the offsets the data map documents for each mode.

Do I have to poll IO-Link parameters, or can the master flag them?

Standard masters do not push ISDU values; your client initiates each read or write. What does arrive unsolicited is the status packed with the cyclic data, such as port communication lost, data invalid or device fault, so program the logic to react to those bits and read the parameter detail explicitly only when a flag is set.

Keep reading

  • Troubleshooting

    Why Micro850 explicit writes fail to SMC EX600 outputs

    Your read works and the write does not because the Micro850 has no EtherNet/IP I/O scanner: its explicit Class 3 messages never supply the Class 1 output ownership the EX600 waits for, so the module holds every channel fail-safe. Set the EX600 documented output force condition, keep service 16 on class 4, instance 150, attribute 3, member 0, and build the write buffer to the documented output image.

    Industrial networks

  • Explainer

    Designing a 1734-AENT Point I/O fiber ring on EtherNet/IP

    Yes — enable IGMP snooping on all seventeen switches: the input data a 1734-AENT adapter produces is multicast, so without it every switch floods the ring and all sixteen adapters reject each other's packets. Full-duplex switching ends collisions but not congestion, each fiber span must meet its own optic's distance limit, and ring recovery must beat four times the RPI.

    Industrial networks

  • How-to

    How to connect PROFINET IO to a CompactLogix through a gateway

    A CompactLogix cannot act as a PROFINET IO controller, so an EtherNet/IP-to-PROFINET gateway does the job: it is an EtherNet/IP adapter to the controller's scanner and the PROFINET IO controller to the field devices. The one reported gotcha on a 1769-L18ER-BB1B with a PLX82-EIP-PNC was byte swapping on both the input and output maps, fixed in the gateway rather than in ladder.

    Industrial networks